Ahhh, August. Got to love it. Summer is in full swing, the weather is generally beautiful (especially here in Chicago), and most of us have either returned from or is on our way to a vacation of one kind or another.
Unfortunately, for many, August is also the start of the much dreaded annual planning and budgeting season. Why is it so dreaded? Because, very few executives and managers enjoy the process of forecasting revenues and expenses for the coming fiscal year, knowing that they will certainly be challenged on their initial forecasts ("Raise your revenue estimates, lower your expense estimates!"). And then they will ultimately be held to and measured against the forecasts throughout the following year, although all kinds of unanticipated things can help or hurt their situation.
Ahhh, the corporate life!
Can we make the forecasting process be less painful? Maybe, but it won't necessarily be easy because much of this involves getting better at understanding and considering uncertainty (or "risk"). As such, there are a number of moving parts, biases and generally accepted behaviors that would need to change. Is it worth it? I think so. In the end the better you are at understanding and considering uncertainty in your forecasts, the better your decisions will get. And better decisions surrounding forecasts often translate into higher odds of success.
To start, let's discuss three major issues with the way many organizations forecast today. First, too much attention is given to the knowable versus the perceived "unknowable." Second, too many organizations manage with a "one number" mentality. And third, at the end of the forecasting process, too few organizations understand what their perceived odds of success are.
Focus more on what we know than what we don't think we can know.
Have you ever noticed how the forecasting of expenses in some organizations is done at a more granular level of detail than revenues? For example, they budget for corporate smart phones down to the exact number in use today, plus the change in the number of phones expected in the coming year. Why? It's not because being off by 100 percent would change their stock prices by 1 cent (it probably wouldn't). It's because the number of phones in use is a knowable figure and the change in phones over the coming year is something that can be controlled. In short, there is little uncertainty, a high ability to control expenses and detailed analysis.
On the other hand, these same organizations likely take a less detailed approach when forecasting revenues. For example, they forecast revenues by product types, customer type or even the combination of both. However, it is rare to find an organization that forecasts revenue down to the exact amount of revenue it expects to generate from each customer, by each product that the customer could buy. Why? Because there is often a lot of uncertainty surrounding the exact amount and types of products customers will buy and what customers eventually end up buying is usually outside of the company's complete control.
Yet if they are off by 100 percent on a revenue line item (up or down), it could have a much more significant impact on the value of the organization. In short, there is a lot of uncertainty, low ability to control revenues and a tendency to have a lot less detailed analysis, especially when compared with the level of detail around expenses.
The challenge is to figure out how to get organizations more comfortable and confident forecasting numbers that have a lot of uncertainty surrounding them and are generally outside of their direct control (most specifically, those line items that could have a significant impact on value).
Change the focus from one target number to the "full distribution." In many organizations, when asking for a forecast for a line item, executives are looking for one number. The processes many organizations use to develop forecast focuses on pegging a target number, where the goal is to hit that number (or better) in the coming year. There is often only limited attention given to the rest of the possible (potentially very wide) distribution. Although convenient, this approach is often very frustrating to those providing the forecasts.
The frequent response when asked for a single-number forecast for a revenue line item is usually along the lines of, "Look, there are a lot of uncertain factors (or "risks") that are outside of my control that could affect where revenues ultimately end up. You can't possibly expect me to know how these factors will play out. To give you a forecast number, the best I can do is make some assumptions and tell you what I think revenue will be given those assumptions."
The issue with this type of approach is that every time someone makes an assumption for a variable, they have also decided to disregard the degree of uncertainty surrounding that exact variable and to ignore information (that is readily available to them) about an influential variable as they move forward in their analysis.
As we all know, it is not usually a great idea to ignore information. Why do organizations do this? Often for the simple reason that you can only populate a single value in a cell in an Excel spreadsheet! Unfortunately, the world we live in is a bit more complex.
"But wait," you are probably saying. "We don't just ask for the 'base case.' We also ask for a 'low case' and 'high case.' "
If so, congratulations. You are one step up on the sophistication ladder. Yet we now have only three points on the distribution. Do you know what the base represents on a percentile (not percentage) basis, and how much lower and how much higher the low case and high case are? Probably not. If you don't know, this could be a bad sign (though not an uncommon phenomena).
Understand the odds of any given game before you place your
bet. If you were to walk up to a gaming table in a casino and there were no explanation of the odds or payoffs of the game, would you bet? Probably not. Making an investment in business should be no different.
Sure, I understand that you can calculate the real odds for a casino game (knowable); whereas, this is far more difficult (and often not possible) in business. Even if you can't calculate the real odds, nevertheless, you should at least calculate your perception of the odds. For instance, it is important to understand if one person perceives the odds of success for an investment to be 75 percent and the rest of the team perceives the odds to be 1 percent. Wouldn't the rest of the team override the one person? (It depends if the one person is the "big cheese" or not.)
This gets us back to why it is important to understand what percentile our base, low and high cases are. Why is this important? Well, if the base is the 25th percentile, then you have a 75 percent probability to achieving that level or higher (i.e., success). If the base is the 50th percentile, then you only have a 50 percent probability of achieving that level or higher. In betting terms, that is a huge difference.
If I were funding your betting as an investor, I would want to make sure that you could articulate your perception of odds on the bets that you are taking.
So what now? Because we now recognize the three major issues with the way many organizations forecast, we can discuss how organizations can improve this situation. In the coming months, we will explore some specific techniques for getting your organization comfortable dealing with key variables that have some degree of uncertainty and are outside of direct control. We will work through a step-by-step approach for collecting a forecast as a distribution instead of as a single number. Finally, we will see if this approach helps us calculate our perceived odds of success.
Ahhh, corporate life!
DAVID M. WONG is director of enterprise risk management at CME Group, the world's largest and most diverse derivatives exchange.
August 23, 2010
Copyright 2010© LRP Publications